If you've been on the internet in the last decade, you've likely witnessed a viral video . If you're anything like me, that video probably left you feeling a mix of impressed, perplexed, and deeply terrified. The bots in question could be [[link]] performing perfectly innocent tasks, or even , and yet there's still an off putting creepiness to watching these human-shaped machines. On the flipside of the uncanny valley, I can watch things like S and feel relatively at ease.
As it turns out, my hypocritical views towards robots are fairly common. For a huge variety of reasons, most people feel pretty uncomfortable when faced with the idea of these metal men, especially when it comes to inviting them into their homes. Despite this, robotics companies seem intent [[link]] on pushing ahead with these bots, despite them often not being as efficient at completing tasks as a purpose built boxy bot.
This contradiction in the trajectory of modern robotics is a question University of Washington robotics professor, Maya Cakmak, has been looking to answer.
Cakmak, alongside students and researchers, conducted several studies on the phenomenon, and detailed the results on . They involved asking people, many of whom already had some robotic assistance in the home, how comfortable they were with humanoid robots vs others for performing certain tasks. The teams found that generally speaking, people were much more comfortable with robots that looked less like people. 
And to be completely honest, yeah, fair enough.
"Not one of them wanted a humanoid," says Cakmak after asking a panel of folks with experience using robotic assistance. "Their concerns ranged from 'it's creepy' to 'it has to be 100 percent safe because I cannot escape it.' One panelist summed it up perfectly: 'Trying to make assistive robots with humanoids would be like trying to make autonomous cars by putting humanoids in the driver’s seat and asking them to drive like a human.'”
As Cakmak explains, the perceived upsides to these robots make sense. If they were proficient enough, a humanoid robot can slot right into any job a person could do, as a sort of multipurpose option. Unfortunately, we aren't really at a point where these upsides are a reality, which instead delivers a visage of robots doing terrible jobs of things like playing sports. Would you really want ?
This is an area Cakmak has a fair amount of expertise in, having an extensive history around working with robots that help people in the home. These are purpose built-bots that are great at specific tasks, and they often look more like something out of the original Lost in Space with boxy bodies and grabber arms. Some even come down from the roof, as most people don't have space in their house for an extra human, let alone one that is likely to fumble around and fall over.
But it's not just bot incompetence that keeps people rightly sceptical, what if they are too good. A Roomba can't exactly chase me down and beat me up like something with the full range of motion of a person can. Whether they're working well or poorly, there are terrifying safety implications around bots that can do so much more than their required function. This is something that participants in the studies brought up, even after being told to assume the bots were perfectly safe.
"Even with our reassurances about safety, people readily imagined hazards: humanoids could trip, stumble, or tip over; they might glitch, run out of battery, or malfunction," [[link]] explains Cakmak
It seems all these experiments have left Cakmak even more unsure about the current billion dollar push towards humanoid robotics. If we know these robots aren't as good as purpose built ones, and 
that people are generally unnerved by them, why is the current vision of future robotics so human shaped? Is it just the virality and sharable nature that's pushing popularity, or do these companies really believe this creepiness is the way forward?
👉👈
1. Best overall:
2. Best budget:
3. Best mid-range:
4. Best high-end:
5. Best AM4 upgrade:
6. Best CPU graphics:
